- The flip canvas was created with swapped width/height for the
vertical flip (as if it were a rotation). A flip is only a mirror:
the canvas now always keeps the original dimensions, for both the
horizontal and the vertical flip.
Files changed:
- web/js/loadImageCrop.js: canvas dimensions no longer swapped
- pyproject.toml: version 1.0.8
- CHANGELOG.md: v1.0.8 entry
- New 'flip' input (None/Horizontal/Vertical, default None).
- Selecting a flip mirrors the image in the browser (canvas), saves it
to input/ as {name}-flipH.png / {name}-flipV.png, and switches the
node's image input to the flipped file.
- The flipped image is used everywhere (preview, mask editor, output);
the crop rectangle is framed on the flipped image (WYSIWYG).
- 'None' reverts to the original image.
- No canvas rendering tricks: the flip is a real file in input/, so it
is consistent across all contexts.
Files changed:
- load_image_crop.py: FLIP_OPTIONS + flip input in INPUT_TYPES
- web/js/loadImageCrop.js: applyFlip() + flip widget binding
- pyproject.toml: version 1.0.7
- CHANGELOG.md: v1.0.7 entry
- README.md: Flip section updated
domg73